IoT-Driven Visual Surveillance: Temporal Masking for Adaptive Motion Compensation in Imaging Technology

Global security is a matter of critical concern that requires adoption of advanced monitoring technologies. Efficient surveillance systems comprise extensive camera networks across large areas to ensure comprehensive coverage. However, the large volume of data generated by these networks poses challenges for traditional storage and computational resources. This paper presents an innovative video compression technique that focuses on optimizing data management in visual surveillance systems by selectively masking temporal information between frames. This technique introduces a specially designed adaptive masking filter, which hides the undetectable motion in video sequences and enhances video compression. The introduced masking technique uses an adaptive masking parameter 'q' to improve frame prediction or to compensate for the masked temporal activity during decoding and achieves over 30% bit-rate reduction compared to the standard video encoding schemes, such as H.264/AVC. Moreover, the introduced technique also reduces the computational demands while keeping the quality of the output. This can be evidenced by a Peak Signal to Noise Ratio (PSNR) of 33.67 dB and a Structural Similarity Index (SSIM) of 92.7% in a traffic video sequence. The proposed technique holds the potential to be used in efficient IoT-driven video surveillance systems to process video frames efficiently without compromising quality.